Subcommittee T-3 of the Advanced Television Systems Committee has been meeting to determine a GCR signal for use in the United States. The GCR signal will be a compromise based from two GCR signals, one using Bessel pulse chirp signals as proposed by U.S. Philips Corp. and one using pseudo noise (PN) sequences as proposed by the David Sarnoff Research Center (DSRC) of Stanford Research Institute. The GCR signals are inserted into selected vertical blanking intervals (VBIs). The GCR signals are used in a television receiver for calculating the adjustable weighting coefficients of a ghost-cancellation filter through which the composite video signals from the video detector are passed to supply a response in which ghosts are suppressed. The weighting coefficients of this ghost-cancellation filter are adjusted so it has a filter characteristic complementary to that of the transmission medium giving rise to the ghosts. The GCR signals can be further used for calculating the adjustable weighting coefficients of an equalization filter connected in cascade with the ghost-cancellation filter, for providing an essentially flat frequency spectrum response over the complete transmission path through the transmitter vestigial-sideband amplitude-modulator, the transmission medium, the television receiver front-end and the cascaded ghost-cancellation and equalization filters.
In the conventional method for cancelling ghosts in a television receiver, the discrete Fourier transform (DFT) of the ghosted GCR signal is divided by the DFT of the non-ghosted GCR signal (which latter OFT is known at the receiver from prior agreement with the transmitter) to obtain as a quotient the DFT transform of the transmission medium giving rise to ghosting; and the inverse DFT of this quotient is then used to define the filter weighing coefficients of a compensating ghost-cancellation filter through which the ghosted composite video signal is passed to obtain a de-ghosted composite video signal. To implement the DFT procedure efficiently, in terms of hardware or of calculations required in software, an integral power of two equal-bandwidth frequency bins are used in the DFT. The distribution of energy in the Philips chirp signal has a frequency spectrum extending continuously across the composite video signal band, in contrast to the DSRC PN sequence in which the distribution of energy does not extend continuously across the composite video signal band, but exhibits nulls in its frequency distribution. Accordingly, when the number of equal-bandwidth frequency bins in the DFT is reduced in order to speed calculation time, more accurate ghost cancellation is obtained with the chirp than with the PN sequence as GCR signal, the inventors observe.
During official testing by the Subcommittee, the DSRC GCR signal has exhibited somewhat better performance in regard to equalization of the passband after ghosting, which some experts including the Philips engineers, attribute to better filter hardware.
Theoretically, equalization calculated over an entire active portion of the VBI, proceeding from the PN sequence, has an accuracy substantially the same as the accuracy available calculating equalization from the chirp signal. The entire length of the Philips chirp signal is needed to have the requisite information to implement equalization over the full composite video signal band, The PN sequence contains pulse transitions each of which transitions has substantially the entire frequency spectrum contained therein. The PN sequence contains many pulse transitions, each of which transitions has component frequencies extending over substantially the entire frequency spectrum. This property of the PN sequence, the inventors observe, permits the calculation of equalization taking samples at a prescribed sampling density only over a limited extent of the GCR signal. Taking samples over only a portion of the GCR signal causes some loss in the accuracy with which equalization can be calculated, particularly under poor signal-to-noise conditions. However, since the number of samples involved in the calculation of weighting coefficients for the equalization filter is reduced, there can be an appreciable increase in the speed with which equalization can be calculated, presuming the calculation is done using an iterative method such as least-mean-squares error reduction. Also, there is reduced complexity, in terms of hardware or of calculations required in software, with regard to calculating the equalization filter weighting coefficients.
The composite GCR signals comprised of chirps and PN sequence signals that have thus far been proposed do not make available both a chirp and a PN sequence during the same VBLI scan line.